Critical Analysis of Hypothesis Tests in Federal Information Processing Standard (140-2)

Area of Science:

• Cryptography
• Statistical Analysis
• Computer Science

Background:

• The FIPS 140-2 standard is crucial for validating cryptographic module security.
• Understanding test dependencies is vital for accurate random number generator (RNG) assessment.
• Existing analyses may not fully capture complex relationships within RNG test suites.

Purpose of the Study:

• To investigate and quantify dependencies between tests in the FIPS 140-2 validation battery.
• To explore both linear and non-linear relationships among RNG tests.
• To provide tools for more rigorous end-user analysis of RNG performance.

Main Methods:

• Pearson's correlation coefficient was employed to identify linear dependencies.
• Mutual information, a novel application in this context, was used to detect non-linear relationships.
• The FIPS 140-2 test battery was reimplemented to facilitate statistical analysis and p-value generation.

Main Results:

• Identified significant linear and non-linear dependencies between various FIPS 140-2 tests.
• The mutual information measure proved effective in uncovering relationships missed by traditional correlation analysis.
• The reimplemented FIPS 140-2 battery provides essential statistics for deeper RNG evaluation.

Conclusions:

• Dependencies within the FIPS 140-2 test battery can impact RNG validation.
• A combined approach using correlation and mutual information offers a more comprehensive understanding of RNG behavior.
• The enhanced FIPS 140-2 analysis tools empower users with greater insight into RNG quality and security.